I ran two equilibration simulations comparing the structures of water-solvated polymer fragments using periodic (NVT) and non-periodic settings of MD simulations with pGFN-FF. According to the results, clearly ordered water layers occurred for the periodic structure, whereas no trace of ordering could be observed for the non-periodic structure (please find snapshots attached). Since the results are a little surprising (unfortunately, no reference for comparison available presently), I would like to ask if there were any relevant parameters for periodic pGFN-FF simulations that one could try adjusting in the case of similar structures.
Best regards,
Hi Osvalds,
My best guess is that this is nothing to do wth pGFN-FF, especially since it’s a molecular system and so the connectivity is pretty unambiguous. I suspect this is just because you’re using NVT for the periodic calculation. This could mean that you are applying pressure to the system which causes it to form an ordered structure, whereas the non-periodic case doesn’t have this applied pressure. There are also the surface effects to consider as well. I’d suggest running NPT with orthorhombic boundary conditions to see what happens. Of course this may depend on where you start from since phase changes between an ordered solid and a disordered liquid hinge on nucleation events and these are tricky things to simulate.
Regards,
Julian
please find a analogous snapshot from an NPT simulation attached. It shows a very similar configuration to the one that was obtained using NVT ensemble. What should be also mentioned here is that by doing NPT equilibration of pure water system about 2x increase of mass density was observed at 1 atm (no ordering changes observed).
Best regards,
Hi Osvalds,
It could be that NPT vs NVT has nothing to do with it, though this is something that could give this effect for some systems. It partly depends on whether you start with an ordered or a disordered system. If the system is initially disordered and goes to ordered then the result is likely to be real. If it starts ordered and only disorders for non-periodic systems that could be a nucleation time/length-scale effect. Another option is that there could be a size-dependent phase transition driven by surface effects. For systems where there are dipoles involved on species (such as water) then the Ewald constraint of no dipole vs the unconstrained dipole for a finite system can make a large difference.
Regards,
Julian
